Datasets:
feature_id int64 0 16.4k | term int64 -1 88.3k | z_score float32 15.6 95.7 ⌀ | auroc float32 0.6 1 ⌀ | label large_stringclasses 351
values | aspect large_stringclasses 9
values | term_by_z int64 -1 88.3k | label_by_z large_stringclasses 319
values | auroc_by_z float32 0.6 1 ⌀ | n_carriers int64 0 4.31k | source_database large_stringclasses 6
values | stable_id large_stringclasses 354
values | term_id large_stringclasses 354
values | held_out_n_positive float64 5 421 ⌀ | n_total float64 2.26k 2.26k ⌀ | held_out_auroc float64 0.4 1 ⌀ | held_out_ap float64 0 1 ⌀ | held_out_prevalence float64 0 0.19 ⌀ | held_out_recall_at_precision_50 float64 0 1 ⌀ | held_out_ap_lift float64 0.88 451 ⌀ | concordance_hits float64 0 20 ⌀ | concordance_drawn float64 5 20 ⌀ | concordance_enrichment float64 0 1.19k ⌀ | concordance_q float64 0 1 ⌀ | biohub_label large_stringlengths 12 66 | assigned bool 2
classes | status large_stringclasses 3
values |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Nudix N-terminal substrate-binding loop | false | no_term_above_threshold | |||
1 | 25,169 | 20.192104 | 0.853027 | pilus | go_cc | 25,169 | pilus | 0.853027 | 41 | go | GO:0009289 | go_cc:go:GO:0009289 | 5 | 2,257 | 0.791563 | 0.086958 | 0.002215 | null | 39.252643 | 0 | 17 | 0 | 1 | PsbO extracellular beta strand signature | true | assigned_and_evaluated |
2 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Acidic Gly-rich capping loops | false | no_term_above_threshold | |||
3 | 88,161 | 57.110443 | 0.68457 | Transmembrane helix | keyword | 88,161 | Transmembrane helix | 0.68457 | 4,197 | keywords | Transmembrane helix | keyword:keywords:Transmembrane helix | 417 | 2,257 | 0.707712 | 0.472334 | 0.184759 | 0.513189 | 2.556493 | 13 | 14 | 5.039814 | 0 | Acidic cytosolic juxtamembrane tails | true | assigned_and_evaluated |
4 | 50,806 | 16.838108 | 0.837402 | Homeobox, conserved site | interpro | 37,071 | Homeodomain | 0.758789 | 52 | ip | IPR017970 | interpro:ip:IPR017970 | null | null | null | null | null | null | null | 0 | 17 | 0 | 1 | Disordered transcriptional effector regions | true | assigned_too_few_test_positives |
5 | 64,594 | 18.268948 | 0.943359 | Zn(2)-C6 fungal-type DNA-binding domain superfamily | interpro | 88,150 | Transcription regulation | 0.643555 | 74 | ip | IPR036864 | interpro:ip:IPR036864 | 8 | 2,257 | 0.990607 | 0.275377 | 0.003545 | null | 77.690787 | 1 | 18 | 13.143959 | 0.119776 | TF regulatory IDRs | true | assigned_and_evaluated |
6 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Transporter TM helices and boundaries | false | no_term_above_threshold | |||
7 | 81,349 | 15.981723 | 0.959473 | 1.10.287.1490 | gene3d | 87,326 | Coiled coil | 0.672852 | 31 | gene3d | 1.10.287.1490 | gene3d:gene3d:1.10.287.1490 | null | null | null | null | null | null | null | 9 | 20 | 161.623193 | 0 | Solvent exposed coiled-coil face | true | assigned_too_few_test_positives |
8 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Lipid transporter transmembrane bundles | false | no_term_above_threshold | |||
9 | 88,058 | 35.459236 | 0.672363 | Signal | keyword | 88,058 | Signal | 0.672363 | 2,747 | keywords | Signal | keyword:keywords:Signal | 206 | 2,257 | 0.665676 | 0.139119 | 0.091272 | null | 1.524231 | 5 | 20 | 1.801239 | 0.213553 | Hydrophobic helical and low-complexity regions | true | assigned_and_evaluated |
10 | 36,449 | 18.616093 | 1 | EamA domain | interpro | 81,858 | Rhodopsin 7-helix transmembrane proteins | 0.893555 | 35 | ip | IPR000620 | interpro:ip:IPR000620 | null | null | null | null | null | null | null | 16 | 19 | 896.556391 | 0 | Aromatic transporter TM helices | true | assigned_too_few_test_positives |
11 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Charged disordered linker/terminal tails | false | no_term_above_threshold | |||
12 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | C-terminal cyclic dinucleotide sensor | false | no_term_above_threshold | |||
13 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | N-terminal β-strand block | false | no_term_above_threshold | |||
14 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Diffuse activation with structured peaks | false | no_term_above_threshold | |||
15 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Disordered low-complexity linkers | false | no_term_above_threshold | |||
16 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Modified peptide core detector | false | no_term_above_threshold | |||
17 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Polyanion-binding amphipathic helical scaffolds | false | no_term_above_threshold | |||
18 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Membrane proteases and catalytic motifs | false | no_term_above_threshold | |||
19 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Catalytic beta-strand in pseudouridine synthases | false | no_term_above_threshold | |||
20 | 88,164 | 20.628948 | 0.768555 | Transposition | keyword | 88,164 | Transposition | 0.768555 | 52 | keywords | Transposition | keyword:keywords:Transposition | 5 | 2,257 | 0.898712 | 0.760443 | 0.002215 | 0.8 | 343.264 | 2 | 16 | 90.885671 | 0.000399 | Transposase C-terminal dimerization helices | true | assigned_and_evaluated |
21 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Charged disordered flanks with motifs | false | no_term_above_threshold | |||
22 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Nucleosidase substrate-binding loops | false | no_term_above_threshold | |||
23 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Extracytosolic ectodomains and His-rich loops | false | no_term_above_threshold | |||
24 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Cytidine deaminase-like domain | false | no_term_above_threshold | |||
25 | 88,058 | 38.671234 | 0.67041 | Signal | keyword | 88,058 | Signal | 0.67041 | 2,747 | keywords | Signal | keyword:keywords:Signal | 206 | 2,257 | 0.660366 | 0.229907 | 0.091272 | 0.106796 | 2.518932 | 12 | 15 | 5.763964 | 0 | Non-cytosolic soluble domains | true | assigned_and_evaluated |
26 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | SLC12 cytoplasmic C-terminal domain | false | no_term_above_threshold | |||
27 | 8,138 | 19.466314 | 0.824219 | translation | go_bp | 88,022 | Ribosomal protein | 0.771484 | 187 | go | GO:0006412 | go_bp:go:GO:0006412 | 41 | 2,257 | 0.63479 | 0.069605 | 0.018166 | 0.02439 | 3.831696 | 0 | 9 | 0 | 1 | NTPase catalytic and coupling residues | true | assigned_and_evaluated |
28 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Jelly-roll/CNBD metal/anion pockets | false | no_term_above_threshold | |||
29 | 47,128 | 17.087891 | 0.977051 | Concanavalin A-like lectin/glucanase domain superfamily | interpro | 47,128 | Concanavalin A-like lectin/glucanase domain superfamily | 0.977051 | 68 | ip | IPR013320 | interpro:ip:IPR013320 | null | null | null | null | null | null | null | 17 | 18 | 242.71001 | 0 | Beta-sandwich/propeller strand cores | true | assigned_too_few_test_positives |
30 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Periplasmic SBP N-terminal ligand-binding lobe | false | no_term_above_threshold | |||
31 | 63,951 | 19.402288 | 0.985352 | Immunoglobulin-like domain superfamily | interpro | 47,529 | Immunoglobulin-like fold | 0.932617 | 77 | ip | IPR036179 | interpro:ip:IPR036179 | 9 | 2,257 | 0.999308 | 0.789269 | 0.003988 | 1 | 197.931065 | 19 | 19 | 161.137838 | 0 | Ig-like YxC disulfide motif | true | assigned_and_evaluated |
32 | 87,882 | 16.713156 | 0.72998 | Palmitate | keyword | 88,058 | Signal | 0.635254 | 251 | keywords | Palmitate | keyword:keywords:Palmitate | 6 | 2,257 | 0.797201 | 0.046946 | 0.002658 | null | 17.659529 | 5 | 18 | 18.504345 | 0.00001 | Extracytoplasmic envelope factors | true | assigned_and_evaluated |
33 | 49,116 | 17.329674 | 0.874023 | Armadillo-type fold | interpro | 43,624 | Homedomain-like superfamily | 0.843262 | 183 | ip | IPR016024 | interpro:ip:IPR016024 | 5 | 2,257 | 0.918606 | 0.016531 | 0.002215 | null | 7.46199 | 1 | 15 | 5.80253 | 0.252889 | N-terminal amphipathic/hydrophobic helix | true | assigned_and_evaluated |
34 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Extended beta-strand/LCR tracts | false | no_term_above_threshold | |||
35 | 88,240 | 15.917836 | 0.61084 | Virulence | keyword | 88,240 | Virulence | 0.61084 | 306 | keywords | Virulence | keyword:keywords:Virulence | 20 | 2,257 | 0.736243 | 0.179888 | 0.008861 | 0.15 | 20.30031 | 2 | 19 | 3.769306 | 0.157825 | Secreted protein mature region | true | assigned_and_evaluated |
36 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Pseudouridine synthase catalytic core | false | no_term_above_threshold | |||
37 | 46,284 | 16.400063 | 0.890137 | FMN-binding split barrel | interpro | 85,712 | 3.40.630.30 | 0.867188 | 26 | ip | IPR012349 | interpro:ip:IPR012349 | null | null | null | null | null | null | null | 0 | 17 | 0 | 1 | IclR effector-binding beta-alpha motif | true | assigned_too_few_test_positives |
38 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Sigma-factor N-terminal IDRs | false | no_term_above_threshold | |||
39 | 64,003 | 16.444174 | 0.882324 | Zinc finger C2H2 superfamily | interpro | 88,150 | Transcription regulation | 0.672363 | 103 | ip | IPR036236 | interpro:ip:IPR036236 | 9 | 2,257 | 0.930308 | 0.213456 | 0.003988 | null | 53.530071 | 9 | 18 | 65.230853 | 0 | Non-core regulatory activation domains | true | assigned_and_evaluated |
40 | 82,029 | 21.456511 | 0.984863 | MFS general substrate transporter like domains | gene3d | 88,161 | Transmembrane helix | 0.667969 | 113 | gene3d | 1.20.1250.20 | gene3d:gene3d:1.20.1250.20 | null | null | null | null | null | null | null | 1 | 17 | 11.063463 | 0.140783 | Generic multi-pass transmembrane helices | true | assigned_too_few_test_positives |
41 | 36,876 | 16.29847 | 1 | Cytochrome P450 | interpro | 87,541 | Heme | 0.666504 | 35 | ip | IPR001128 | interpro:ip:IPR001128 | 9 | 2,257 | 1 | 1 | 0.003988 | 1 | 250.777778 | 15 | 15 | 389.679739 | 0 | Cytochrome P450 catalytic fold | true | assigned_and_evaluated |
42 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Short conserved ligand-binding patch | false | no_term_above_threshold | |||
43 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Kinase and transferase catalytic cores | false | no_term_above_threshold | |||
44 | 63,960 | 32.794453 | 0.928223 | FAD/NAD(P)-binding domain superfamily | interpro | 63,960 | FAD/NAD(P)-binding domain superfamily | 0.928223 | 72 | ip | IPR036188 | interpro:ip:IPR036188 | null | null | null | null | null | null | null | 15 | 15 | 193.574675 | 0 | N-terminal FAD-binding segment | true | assigned_too_few_test_positives |
45 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Extended noncatalytic interaction segments | false | no_term_above_threshold | |||
46 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | ScpA/NSE4/EID3 mid-chain activation | false | no_term_above_threshold | |||
47 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Broad conserved-motif activation | false | no_term_above_threshold | |||
48 | 81,739 | 15.974072 | 0.959473 | DNA helicase RuvA subunit, C-terminal domain | gene3d | 81,739 | DNA helicase RuvA subunit, C-terminal domain | 0.959473 | 27 | gene3d | 1.10.8.10 | gene3d:gene3d:1.10.8.10 | 6 | 2,257 | 0.999704 | 0.910714 | 0.002658 | 1 | 342.580357 | 11 | 18 | 278.130195 | 0 | C-terminal disordered low-complexity tails | true | assigned_and_evaluated |
49 | 37,097 | 19.431948 | 0.998535 | Cro/C1-type helix-turn-helix domain | interpro | 37,071 | Homeodomain | 0.945801 | 49 | ip | IPR001387 | interpro:ip:IPR001387 | null | null | null | null | null | null | null | 18 | 19 | 532.858987 | 0 | HTH/LysM hydrophobic packing | true | assigned_too_few_test_positives |
50 | 50,338 | 17.492237 | 0.825195 | GPCR, rhodopsin-like, 7TM | interpro | 50,338 | GPCR, rhodopsin-like, 7TM | 0.825195 | 27 | ip | IPR017452 | interpro:ip:IPR017452 | null | null | null | null | null | null | null | 0 | 19 | 0 | 1 | Asn-biased coiled-coil docking helices | true | assigned_too_few_test_positives |
51 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | uS2 C-terminal helix motif | false | no_term_above_threshold | |||
52 | 38,873 | 16.893593 | 0.755371 | AAA+ ATPase domain | interpro | 58,239 | P-loop containing nucleoside triphosphate hydrolase | 0.631836 | 60 | ip | IPR003593 | interpro:ip:IPR003593 | 5 | 2,257 | 0.998224 | 0.635965 | 0.002215 | 0.8 | 287.074561 | 15 | 16 | 160.616918 | 0 | ABC NBD LSGGQ signature | true | assigned_and_evaluated |
53 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Two-metal phosphodiesterase/nuclease cores | false | no_term_above_threshold | |||
54 | 63,831 | 31.773422 | 0.929199 | F-box-like domain superfamily | interpro | 63,831 | F-box-like domain superfamily | 0.929199 | 115 | ip | IPR036047 | interpro:ip:IPR036047 | 6 | 2,257 | 0.964238 | 0.039416 | 0.002658 | null | 14.827121 | 0 | 14 | 0 | 1 | REC β3–α2 and H-box | true | assigned_and_evaluated |
55 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Enzyme N-terminal extensions | false | no_term_above_threshold | |||
56 | 36,535 | 18.96195 | 0.871094 | Protein kinase domain | interpro | 58,239 | P-loop containing nucleoside triphosphate hydrolase | 0.641113 | 86 | ip | IPR000719 | interpro:ip:IPR000719 | 10 | 2,257 | 0.891811 | 0.217809 | 0.004431 | null | 49.159451 | 0 | 16 | 0 | 1 | ABC NBD beta1/A-loop | true | assigned_and_evaluated |
57 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Exposed aromatic glycan-binding loops | false | no_term_above_threshold | |||
58 | 88,263 | 18.247887 | 0.70459 | Zinc-finger | keyword | 88,263 | Zinc-finger | 0.70459 | 529 | keywords | Zinc-finger | keyword:keywords:Zinc-finger | 33 | 2,257 | 0.796599 | 0.095002 | 0.014621 | null | 6.497538 | 3 | 18 | 4.343022 | 0.049818 | Modular globular binding domains | true | assigned_and_evaluated |
59 | 36,535 | 18.636656 | 0.975586 | Protein kinase domain | interpro | 85,656 | alpha/beta hydrolase | 0.924316 | 86 | ip | IPR000719 | interpro:ip:IPR000719 | 10 | 2,257 | 0.989809 | 0.742644 | 0.004431 | 0.8 | 167.614755 | 16 | 16 | 89.790663 | 0 | Kinase catalytic core motifs | true | assigned_and_evaluated |
60 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Extended electrostatic interaction surfaces | false | no_term_above_threshold | |||
61 | 38,654 | 17.740204 | 0.876465 | B3 DNA binding domain | interpro | 38,654 | B3 DNA binding domain | 0.876465 | 35 | ip | IPR003340 | interpro:ip:IPR003340 | null | null | null | null | null | null | null | 0 | 16 | 0 | 1 | Acidic N-terminal activation IDRs | true | assigned_too_few_test_positives |
62 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Extracellular vestibule pore loops | false | no_term_above_threshold | |||
63 | 69,819 | 19.020533 | 0.737305 | ATPase, nucleotide binding domain | interpro | 84,753 | 3.30.420.40 | 0.731934 | 35 | ip | IPR043129 | interpro:ip:IPR043129 | null | null | null | null | null | null | null | 13 | 15 | 363.884038 | 0 | Hsp70 substrate-binding lobe activation | true | assigned_too_few_test_positives |
64 | 88,009 | 30.916613 | 0.725586 | Repeat | keyword | 88,009 | Repeat | 0.725586 | 1,599 | keywords | Repeat | keyword:keywords:Repeat | 164 | 2,257 | 0.840776 | 0.321138 | 0.072663 | 0.231707 | 4.419567 | 12 | 14 | 7.599065 | 0 | KSRPYF coiled-coil motif | true | assigned_and_evaluated |
65 | 82,029 | 21.453115 | 0.996582 | MFS general substrate transporter like domains | gene3d | 88,161 | Transmembrane helix | 0.646484 | 113 | gene3d | 1.20.1250.20 | gene3d:gene3d:1.20.1250.20 | null | null | null | null | null | null | null | 12 | 18 | 125.38591 | 0 | Transporter TM hydrophobic hotspot detector | true | assigned_too_few_test_positives |
66 | 49,061 | 15.694553 | 0.785156 | WD40/YVTN repeat-like-containing domain superfamily | interpro | 49,061 | WD40/YVTN repeat-like-containing domain superfamily | 0.785156 | 157 | ip | IPR015943 | interpro:ip:IPR015943 | 28 | 2,257 | 0.872004 | 0.059542 | 0.012406 | null | 4.799492 | 0 | 16 | 0 | 1 | Beta-propeller blade motifs | true | assigned_and_evaluated |
67 | 47,529 | 25.497551 | 0.974121 | Immunoglobulin-like fold | interpro | 47,529 | Immunoglobulin-like fold | 0.974121 | 157 | ip | IPR013783 | interpro:ip:IPR013783 | 16 | 2,257 | 0.994436 | 0.627906 | 0.007089 | 0.6875 | 88.573972 | 17 | 17 | 77.329442 | 0 | Repeated Ig/FN3 loop hydrophobics | true | assigned_and_evaluated |
68 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Transmembrane helix termini and loops | false | no_term_above_threshold | |||
69 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Periodic aromatic/Gly beta-repeat detector | false | no_term_above_threshold | |||
70 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Extracellular anchoring/assembly; GH16 motif | false | no_term_above_threshold | |||
71 | 61,392 | 22.668465 | 0.998535 | Leucine-rich repeat domain superfamily | interpro | 88,009 | Repeat | 0.610352 | 80 | ip | IPR032675 | interpro:ip:IPR032675 | 7 | 2,257 | 0.99873 | 0.627417 | 0.003101 | 1 | 202.297176 | 20 | 20 | 167.474719 | 0 | LRR convex-face helix detector | true | assigned_and_evaluated |
72 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | sPLA2 N-terminal disulfide/Ca2+ module | false | no_term_above_threshold | |||
73 | 38,654 | 22.004837 | 0.905762 | B3 DNA binding domain | interpro | 48,767 | DNA-binding pseudobarrel domain superfamily | 0.885742 | 35 | ip | IPR003340 | interpro:ip:IPR003340 | null | null | null | null | null | null | null | 0 | 12 | 0 | 1 | Pepsin-like aspartyl proteases | true | assigned_too_few_test_positives |
74 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Radical SAM downstream aromatic/charged segment | false | no_term_above_threshold | |||
75 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Aromatic-capped flexible loops | false | no_term_above_threshold | |||
76 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Kinase TGD catalytic loop | false | no_term_above_threshold | |||
77 | 87,326 | 35.122238 | 0.733398 | Coiled coil | keyword | 87,326 | Coiled coil | 0.733398 | 1,122 | keywords | Coiled coil | keyword:keywords:Coiled coil | 89 | 2,257 | 0.76824 | 0.38239 | 0.039433 | 0.404494 | 9.697238 | 17 | 19 | 11.28042 | 0 | Coiled-coil heptad register | true | assigned_and_evaluated |
78 | 36,776 | 18.12604 | 0.997559 | SANT/Myb domain | interpro | 80,926 | Homeodomain-like | 0.891602 | 45 | ip | IPR001005 | interpro:ip:IPR001005 | 6 | 2,257 | 0.998815 | 0.616071 | 0.002658 | 1 | 231.745536 | 0 | 19 | 0 | 1 | HTH-like DNA-contacting turn | true | assigned_and_evaluated |
79 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Pervasive N-terminal nucleic-acid activation | false | no_term_above_threshold | |||
80 | 27,735 | 17.149786 | 0.708984 | DNA-binding transcription factor activity, RNA polymerase II-specific | go_mf | 27,735 | DNA-binding transcription factor activity, RNA polymerase II-specific | 0.708984 | 261 | go | GO:0000981 | go_mf:go:GO:0000981 | 26 | 2,257 | 0.677947 | 0.094985 | 0.01152 | 0.115385 | 8.24546 | 15 | 16 | 42.408716 | 0 | Disordered low-complexity activation segments | true | assigned_and_evaluated |
81 | 47,531 | 17.095268 | 0.86084 | Aldolase-type TIM barrel | interpro | 47,531 | Aldolase-type TIM barrel | 0.86084 | 52 | ip | IPR013785 | interpro:ip:IPR013785 | null | null | null | null | null | null | null | 12 | 12 | 252.631356 | 0 | Dus N-terminal TIM-barrel segment | true | assigned_too_few_test_positives |
82 | 87,856 | 17.251034 | 0.750977 | Nucleotidyltransferase | keyword | 87,856 | Nucleotidyltransferase | 0.750977 | 166 | keywords | Nucleotidyltransferase | keyword:keywords:Nucleotidyltransferase | 28 | 2,257 | 0.58073 | 0.081413 | 0.012406 | 0.107143 | 6.562473 | 14 | 14 | 84.68892 | 0 | Right-hand polymerase catalytic core | true | assigned_and_evaluated |
83 | 63,774 | 16.153095 | 0.999512 | RNA-binding domain superfamily | interpro | 46,581 | Nucleotide-binding alpha-beta plait domain superfamily | 0.998535 | 69 | ip | IPR035979 | interpro:ip:IPR035979 | null | null | null | null | null | null | null | 6 | 15 | 73.154601 | 0 | Exposed aromatic stacking patches | true | assigned_too_few_test_positives |
84 | 82,107 | 15.919426 | 0.629883 | 1.20.1280.50 | gene3d | 63,831 | F-box-like domain superfamily | 0.612305 | 81 | gene3d | 1.20.1280.50 | gene3d:gene3d:1.20.1280.50 | null | null | null | null | null | null | null | 2 | 18 | 36.199757 | 0.002424 | RHS/YD β-strand repeat markers | true | assigned_too_few_test_positives |
85 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Histidine-kinase/TatB cytoplasmic regions | false | no_term_above_threshold | |||
86 | 36,886 | 23.416454 | 0.959473 | Zn(2)Cys(6) fungal-type DNA-binding domain | interpro | 36,886 | Zn(2)Cys(6) fungal-type DNA-binding domain | 0.959473 | 75 | ip | IPR001138 | interpro:ip:IPR001138 | 8 | 2,257 | 0.932414 | 0.42112 | 0.003545 | 0.5 | 118.808464 | 0 | 17 | 0 | 1 | GPCR/MFS TM core motifs | true | assigned_and_evaluated |
87 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Rod-like coiled-coil scaffolds | false | no_term_above_threshold | |||
88 | 85,762 | 18.258623 | 0.685059 | FAD/NAD(P)-binding domain | gene3d | 63,960 | FAD/NAD(P)-binding domain superfamily | 0.67334 | 65 | gene3d | 3.50.50.60 | gene3d:gene3d:3.50.50.60 | null | null | null | null | null | null | null | 14 | 14 | 205.589655 | 0 | FAD oxidoreductase capping motif | true | assigned_too_few_test_positives |
89 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Charged low-complexity IDR tails | false | no_term_above_threshold | |||
90 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Noncatalytic peripheral amphipathic helices/loops | false | no_term_above_threshold | |||
91 | 64,620 | 19.361362 | 0.983398 | Histidine kinase/HSP90-like ATPase superfamily | interpro | 64,620 | Histidine kinase/HSP90-like ATPase superfamily | 0.983398 | 31 | ip | IPR036890 | interpro:ip:IPR036890 | null | null | null | null | null | null | null | 18 | 18 | 324.027174 | 0 | HATPase_c F-box motif | true | assigned_too_few_test_positives |
92 | 36,535 | 20.795782 | 0.952637 | Protein kinase domain | interpro | 81,685 | Transferase(Phosphotransferase) domain 1 | 0.939941 | 86 | ip | IPR000719 | interpro:ip:IPR000719 | 10 | 2,257 | 0.947419 | 0.890443 | 0.004431 | 0.9 | 200.973 | 13 | 13 | 89.790663 | 0 | Kinase P+1 loop LTGxPPF/Y | true | assigned_and_evaluated |
93 | 87,109 | 15.849942 | 0.793945 | ATP synthesis | keyword | 87,326 | Coiled coil | 0.623535 | 92 | keywords | ATP synthesis | keyword:keywords:ATP synthesis | 24 | 2,257 | 0.930587 | 0.167971 | 0.010634 | 0.083333 | 15.796259 | 9 | 14 | 185.158385 | 0 | Long coiled-coil stator scaffolds | true | assigned_and_evaluated |
94 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Gly/Ser/Pro-rich junction loops | false | no_term_above_threshold | |||
95 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Acidic N-terminal IMPase lobe | false | no_term_above_threshold | |||
96 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Phosphorylation-prone disordered linkers | false | no_term_above_threshold | |||
97 | 87,326 | 20.325161 | 0.650391 | Coiled coil | keyword | 87,326 | Coiled coil | 0.650391 | 1,122 | keywords | Coiled coil | keyword:keywords:Coiled coil | 89 | 2,257 | 0.670094 | 0.089494 | 0.039433 | null | 2.269534 | 7 | 19 | 4.644879 | 0.00075 | Coiled-coil oligomerization interface | true | assigned_and_evaluated |
98 | 49,116 | 23.934456 | 0.956543 | Armadillo-type fold | interpro | 49,116 | Armadillo-type fold | 0.956543 | 183 | ip | IPR016024 | interpro:ip:IPR016024 | 5 | 2,257 | 0.997425 | 0.32359 | 0.002215 | null | 146.06841 | 16 | 18 | 77.367072 | 0 | ARM/HEAT/TPR alpha-solenoid repeats | true | assigned_and_evaluated |
99 | -1 | null | null | -1 | null | 0 | null | null | null | null | null | null | null | null | null | null | null | null | null | null | Disordered regions and collagen repeats | false | no_term_above_threshold |
Vocabulary interpretations of ESMC-6B SAE features
One row for every one of the 16,384 features of
biohub/ESMC-6B-sae-layer60-k64-codebook16384, giving the protein annotation vocabulary term
that best identifies what the feature detects, together with how well that identification holds on
proteins the assignment never saw.
This is the counterpart to biohub/ESMC-SAE-Features, produced without a language model. Where that
release gives a free-text hypothesis per feature, this gives a controlled-vocabulary term plus a
falsifiable prediction: which held-out proteins the feature should fire on, and how well it does.
The Hub exposes separate global and local configurations. global is the default; each
configuration contains one train split backed by the correspondingly named Parquet table below.
What the numbers mean
Every measurement here comes from one of four protein sets that never overlap, because splits are assigned to whole MMseqs2 clusters at 30% identity and 50% coverage:
| Set | Role |
|---|---|
| Discovery, 26,725 train-split cluster representatives | fits the assignment |
| Validation, 2,402 proteins in 1,582 clusters | chooses the ranking rule and support floor, nothing else |
| Test, 2,257 proteins in 1,513 clusters | evaluated once against the frozen choice |
| Biohub's released exemplars | never seen by any of the three; every cluster containing one is removed from all of them |
A feature receives a term only if the term clears the largest feature-wise best-term association
observed in four cluster-block permutation draws over the full 16,384-feature family. That
guardrail is z = 15.56; a nominal Benjamini-Hochberg cut over the same tests would sit at 4.36.
Four permutation maxima cannot calibrate a low family-wise error rate, so z = 15.56 is a coarse,
uncalibrated guardrail rather than a calibrated significance threshold. Among terms clearing it,
the term with the highest AUROC wins, subject to a floor of 25 carriers.
Columns
feature_interpretations.parquet, 16,384 rows
| Column | Meaning |
|---|---|
feature_id |
index into the codebook, 0 to 16,383 |
status |
assigned_and_evaluated, assigned_too_few_test_positives, or no_term_above_threshold |
term_id |
V1-derived display composite, aspect:source_database:stable_id; it is not a canonical av::... identifier or a direct key into the 2026_02 term dictionary |
stable_id, source_database |
the parts term_id is composed from |
term |
raw V1 integer token for assigned rows; unassigned rows use the sentinel -1 |
n_total, assigned |
held-out test population size, 2,257, the denominator of held_out_prevalence and non-null only for the 4,470 evaluated features; and whether a term cleared every threshold |
label, aspect |
human-readable term name and which ontology it came from |
z_score, auroc, n_carriers |
discovery-set association strength, effect size, and support |
term_by_z, label_by_z, auroc_by_z |
the alternative ranking, for comparison |
held_out_auroc |
AUROC on the cluster-disjoint test split |
held_out_ap, held_out_ap_lift, held_out_prevalence |
average precision, and its lift over the base rate |
held_out_recall_at_precision_50 |
null for two different reasons: 1,828 evaluated features never reach precision 0.5, and 11,914 were not evaluated at all |
held_out_n_positive |
test-split carriers; null where the feature was not evaluated |
concordance_hits, concordance_drawn, concordance_enrichment, concordance_q |
agreement with Biohub's own 20 top-activating proteins for that feature |
biohub_label |
the released agent label, for side-by-side reading |
Do not join term_id directly to the 2026_02 term dictionary. This archival file's display
composite predates the current av::aspect::source_database::stable_accession identifier grammar.
For assigned global rows, migrate the raw V1 term through the old_token key in the migration
configuration of Synthyra/SwissProt-Annotation-Vocabulary at immutable revision
045af7cdd38e83462ebb10882b7506deff21e658. Only exact and replaced outcomes provide an
automatic current new_term_id and new_token; every other status requires the handling described
by that release. Exclude unassigned rows before the join because their term value is the -1
sentinel, not a biological token.
from datasets import load_dataset
features = load_dataset(
"Synthyra/ESMC-6B-SAE-Annotation-Vocabulary-Features",
"global",
split="train",
revision="03472f5ddce0cb5608f5c5b5e27b2f4bdc0a273c",
token=False,
).to_pandas()
assigned = features.loc[features["assigned"].astype(bool)].copy()
assert assigned["term"].ge(0).all()
migration = load_dataset(
"Synthyra/SwissProt-Annotation-Vocabulary",
"migration",
split="migration",
revision="045af7cdd38e83462ebb10882b7506deff21e658",
token=False,
).to_pandas()
migration = migration.rename(columns={"status": "migration_status"})
current = assigned.merge(
migration[["old_token", "migration_status", "new_term_id", "new_token"]],
left_on="term",
right_on="old_token",
how="left",
validate="many_to_one",
indicator=True,
)
assert current["_merge"].eq("both").all()
current["automatic_current_mapping"] = current["migration_status"].isin(
["exact", "replaced"]
)
At these pinned revisions, all 8,226 assigned feature rows join to the migration table: 5,482 are
exact, 1,363 are replaced, and 1,381 are removed/unmapped. The first two groups, 6,845 rows,
have an automatic current mapping.
local_associations.parquet
Residue-level associations between features and annotated spans, tested within each protein against a hypergeometric null and combined across proteins by Cochran-Mantel-Haenszel.
| Column | Meaning |
|---|---|
feature_id, local_term, local_term_id, local_stable_id |
the V1-derived local pair. The local vocabulary is not the global one: do not join these identifiers or integers directly to either the global table or the 2026_02 release |
label, aspect |
the local term's name and which UniProt feature key it came from |
z_score, enrichment, observed, expected |
inside-span activity against the within-protein expectation |
n_strata |
contributing proteins |
above_positional_null_max |
the column that matters, see below |
positional_null_max_z |
the guardrail value that flag is taken against, 61.96 |
The file ships 88,717 pairs above a loose z = 4. That cut is not a finding: reshuffling every span
to a random position inside its own protein, preserving length and count, still puts 6.0% of pairs
above it. Filter on above_positional_null_max for the 2,907 pairs over 1,985 features that exceed
every value the positional null produced across 507,098 reshuffled pairs. That is an observed
guardrail, the largest value the null happened to reach, not a calibrated error rate.
How well the interpretations hold
| Features with a term | 8,226 of 16,384 |
| Distinct terms used | 354 |
| Features actually evaluated held-out | 4,470 of the 8,226 |
| Median held-out AUROC | 0.889, cluster-bootstrap 95% interval 0.872 to 0.913 |
| Held-out AUROC above 0.8 | 63.2% of the 4,470 evaluated, 34.3% of the 8,226 assigned |
| Median precision lift over base rate | 27.5x |
| Concordance with Biohub's exemplars | 29.2x their Swiss-Prot base rate, 58.2% significant at q < 0.01, over the 8,224 features that could be scored |
Read the denominators. A feature is evaluated held-out only if its term has at least five carriers among the 2,257 test proteins, which 4,470 of the 8,226 assigned features do. The other 3,756 are not failures; the test split was too small to score them, and their held-out columns are null. Concordance is over 8,224 features, two assigned features having no scoreable exemplar set.
Limitations, stated rather than buried
- This is concordance, not accuracy. Biohub publishes no evaluation of description quality and its own tutorial calls the descriptions "automatically generated hypotheses". There is no ground truth here to be right about, and none is claimed.
- 8,226 features, 354 terms. Roughly 10 features share each term at the median. Those features are not duplicates, they nominate largely different proteins with median pairwise Jaccard 0.072, but the number of distinct nameable concepts is bounded by the discovery set, not by the codebook.
- A vocabulary term is not a description. Where a feature encodes something the vocabulary
cannot name, the method returns the nearest available term or nothing. Nine of the ten strongest
assignments resolve distinct parts of a bacterial signal peptide, and all nine receive the
single keyword
Signal, which 120 features carry in total. - The test split is cluster-disjoint but not pristine. It has been evaluated under several run configurations. What is claimed is narrower and is what the protocol enforces: the ranking rule and support floor were chosen on validation alone, and no configuration was selected on test performance.
- The association ran against the previous vocabulary, which still contains Gene3D. The table contains 1,378 Gene3D assignments across 42 Gene3D terms. Of those assigned features, 556 were evaluated on the held-out split and 822 had too few held-out positives to evaluate. Re-running with Gene3D withheld gives 8,020 assignments at held-out AUROC 0.884 against 8,226 at 0.889, so the difference is small, but it is a difference. Gene3D is excluded from the released 2026_02 annotation vocabulary.
- The identifier columns use V1-derived grammar. None of the 8,226 assigned global
term_idvalues or 88,717local_term_idvalues uses the current canonicalav::...grammar. Assigned global raw tokens migrate throughmigration.old_tokenas shown above; unassigned global rows useterm = -1. The local tokens span 88,286 through 90,072, outside the migration table's 0-through-88,280 key range. No released automatic mapping exists for these description-specific local hypotheses, although a future manual semantic mapping remains possible. - Provenance is partial.
provenance.jsonrecords digests of the analysis inputs, the analysis modules, and the run's own outputs, computed over at most the first 1 GiB of any file. It does not contain digests of the six files shipped here, and the module digests describe the code as it stood at the run, not necessarily the current repository. - The
biohub_labelcolumn is redistributed from Biohub's released feature table and is governed by that release's terms, not by the license on this dataset. Drop the column if that matters for your use. - Activations are recomputed, not the released ones. Fused attention and layer-norm kernels were unavailable, so values differ numerically. The released exemplars sit at median percentile 0.99986 of our activation distribution for their feature; the same check against an unrelated codebook gives 0.464.
Provenance
provenance.json carries SHA-256 digests of every input, every analysis module, the resolved
options, and every output file, plus the seed and permutation count. Analysis code and the report
that describes the method are in the accompanying repository.
Citation
If you use this table, please also cite Biohub's ESMC and SAE releases, whose feature table and exemplar proteins make the comparison in this work possible.
- Downloads last month
- 75